New cut-off values for ferritin and soluble transferrinreceptor for the assessment of iron deficiency inchildren in a high infection pressure area

K S Phiri,1 J C J Calis,1,2 A Siyasiya,1 I Bates,3 B Brabin,3 M Boele van Hensbroek1,2,3

1 Malawi–Liverpool–WellcomeTrust Clinical ResearchProgramme, College ofMedicine, Blantyre, Malawi;2 Emma Children’s HospitalAMC, University of Amsterdam,the Netherlands; 3 LiverpoolSchool of Tropical Medicine,Liverpool, UK

Correspondence to:Dr K Phiri, Malawi–Liverpool–Wellcome Trust ClinicalResearch Programme, PO Box30096, Blantyre 3, Malawi;kamijaphiri@gmail.com

Accepted 18 August 2009

This paper is freely availableonline under the BMJ Journalsunlocked scheme, see http://jcp.bmj.com/info/unlocked.dtl

ABSTRACTBackground: Due to the potential risk of iron supple-mentation in iron replete children, it is important toproperly identify children who may require iron supple-mentation. However, assessment of the iron status hasproven to be difficult, especially in children living in areaswith high infection pressure (including malaria).Aims and Methods: Biochemical iron markers werecompared to bone marrow iron findings in 381 Malawianchildren with severe anaemia.Results: Soluble transferrin receptor/log ferritin (TfR-Findex), using a cut-off of 5.6, best predicted bone marrowiron stores deficiency (sensitivity 74%, specificity 73%,accuracy 73%). In order to improve the diagnosticaccuracy of ferritin or sTfR as a stand-alone marker, thenormal cut-off value needed to be increased by 810% and83% respectively. Mean cell haemoglobin concentration(MCHC), using a cut-off of 32.1 g/dl, had a sensitivity of67% and specificity of 64% for detecting iron storesdeficiency.Conclusion: TfR-F index incorporated the high sensitivityof sTfR, a proxy for cellular iron need, and the highspecificity of ferritin, a proxy for iron stores. In areas witha high infection pressure, the TfR-F index best predictediron deficiency. However, in settings where diagnostictests are limited, MCHC may be an acceptable alternativescreening test.

The examination of stained aspirates of bonemarrow for haemosiderin has been considered the‘‘gold standard’’ as a method for evaluation of ironstatus.1 This technique is invasive and thereforenot suitable for screening. There are currentlyseveral laboratory assays available for assessing theiron status in children. These are widely used inclinics and include ferritin, serum iron, serumtransferrin, total iron binding capacity (TIBC)and mean cell volume (MCV). However, theseiron markers are considerably altered by inflamma-tion, which limits their applicability, especially inareas with a high infection pressure. Surprisingly, itis still unclear to what extent adjustment of therecommended cut-off values for these iron markersis required in order to improve their diagnosticefficiency. To date there are no studies which havevalidated these iron markers against the ‘‘goldstandard’’ (bone marrow iron content) in childrenliving in areas endemic for malaria and othercommon infective conditions. In areas with alimited infection pressure, soluble transferrinreceptor (sTfR) has been shown to be a promisingnew tool for the diagnosis of deficiency of ironstores.2

Due to a recent finding in Tanzania of anincreased mortality in iron replete children receiv-ing iron supplementation, there is an urgent needto be able to target iron therapy and prophylaxisprogrammes on the children with proven irondeficiency.3 To be able to do this there is a need fornon-invasive and sensitive tests that distinguishiron stores deficiency from functional iron defi-ciency which is associated with anaemia ofinflammation. We have evaluated the diagnosticaccuracy of various iron markers against bonemarrow iron assessment in children residing in anarea of high infection pressure.

METHODSThis study formed part of a large case–controlstudy investigating the aetiology of severe anaemiaamong Malawian children described in detailelsewhere.4 In summary, the study was conductedbetween July 2002 and July 2004 in malariaendemic areas of Blantyre and Chikwawa.Children aged 6–59 months, presenting to hospitalwith severe anaemia (Hb ,5 g/dl) were enrolled ascases. Written informed consent was obtainedfrom the guardians of the children, and the studywas approved by the ethics committees of theUniversity of Malawi and the Liverpool School ofTropical Medicine, UK.

Clinical assessment and managementChildren had a detailed medical examinationperformed before collecting a sample of venousblood. Under local anaesthesia, a bone marrowaspiration was performed from the anterior orposterior iliac crest. All bone marrows wereperformed in the presence of the child’s guardianand she/he was allowed to withdraw consent atany time during the procedure. All childrenreceived a blood transfusion and were managedaccording to standard hospital protocols.4

Laboratory methodsOn recruitment, haematological indices (includinghaemoglobin (Hb), mean cell volume (MCV) andmean cell haemoglobin concentration (MCHC)) weredetermined using the HemoCue B-Haemoglobin(HemoCue AB, Angelholm, Sweden) and a coultercounter analyser (Beckman Coulter, Durban, SouthAfrica). Ferritin, serum iron and serum transferrinwere also assayed (Roche Diagnostics, Switzerland).sTfR was measured using an enzyme immunoassay(Ramco Laboratories, Texas, USA). Blood was cul-tured for 5 and 56 days for the presence of routinepathogens and mycobacteria, respectively, using an

Original article

J Clin Pathol 2009;62:1103–1106. doi:10.1136/jcp.2009.066498 1103

automated BacT/Alert system (BioMerieux Industry,Missouri, USA).

Definitions and cut-off valuesMalaria was defined as presence of Plasmodium falciparumasexual parasites in blood. HIV testing was done using tworapid tests according to WHO guidelines and discordant resultswere resolved by PCR.5

Internationally accepted cut-off values for biochemical ironmarkers used in this analysis were as follows: ferritin ,30 mg/l6;serum iron ,3.6 mmol/l; serum transferrin .3.6 g/l; TIBC .

72 mmol/l (laboratory reference values); transferrin saturation,16%6; sTfR .8.3 mg/l (test kit reference value); MCV ,67 fl(,2 years old) and ,73 fl (2–5 years old); MCHC ,32 g/l.6

Transferrin-ferritin (TfR-F) index was defined as [sTfR 4 logferritin].7 A TfR-F index value of .5.6 (using sTfR of .8.3 mg/l andferritin of ,30 mg/l) was used to define deficiency of iron stores.8 C-reactive protein (CRP) of .10 mg/ml was defined as increased.9

Bone marrow aspirates were prepared and stained using acommercial kit and according to manufacturer guidelines(HematoGnost Fe, Darmstadt, Germany). Smears were gradedfor iron and defined as positive when the fragment iron was,2.8 Bone marrow iron assessment was used as the goldstandard for diagnosis of iron stores deficiency.

Statistical analysis for sensitivity, specificity, positive pre-dictive value (PPV), negative predictive value (NPV) andaccuracy for individual iron markers was calculated using

SPSS V.11.0. Iron markers having either a sensitivity orspecificity ,20% were classified as poor predictors of ironstores deficiency and were dropped from further analysis. Forthe remaining iron markers, receiver operating characteristics(ROC) curves for identifying the optimal cut-offs for bestidentifying iron stores deficiency were constructed, and thecorresponding areas under the curve (AUCROC) for all themarkers were compared. A new cut-off for each iron markerwhich provided maximal sensitivity and specificity wasdetermined from ROC curves.10 KSP analysed the data but allothers had access to the primary dataset.

RESULTSA total of 381 children were recruited with an average age of 1.7years (SD 1.1); 46.7% (178/381) were male (table 1). Sixty percent of severely anaemic children had malaria parasites in theirblood; CRP was raised in 89%. Ferritin levels were increased,with a mean concentration of 729.2 (1528.1) mg/l (table 2). Theproportions of children that were iron deficient ranged from 1%using TIBC to 97.5% using serum transferrin.

Table 3 shows sensitivity and specificity of various ironmarkers. Poor markers that were dropped from further analysisincluded serum transferrin, TIBC and MCV, which had asensitivity of 0%, 0% and 4% (,2 years) and 17% (>2 years),respectively. Ferritin, serum iron and transferrin saturation had

Table 1 Baseline characteristics of cases

Characteristic Result

Recruited n = 381

Age (months)* 20.4 (12.8) [381]

Hb (g/dl)* 3.6 (0.8) [381]

CRP (mg/ml)* 11.1 (8.5) [346]

Male 178/381 (46.7%)

Wasted{ 52/330 (15.8%)

Stunted{ 176/331 (53.2%)

Malaria 226/380 (59.5%)

HIV 33/345 (9.6%)

Bacteraemia 52/259 (20.1%)

*Mean (SD) [total number].{Z-score ,22 weight-for-height.{Z-score ,22 height-for-age.CRP, C-reactive protein; Hb, haemoglobin.

Table 2 Mean iron marker and the proportion of children classified asiron stores deficient using internationally accepted cut-off values

Iron marker Mean (SD) Normal levelsProportion irondeficient (%)

Ferritin (mg/l) 729.2 (1528.1) 30–300 12.7

sTfR (mg/ml) 17.4 (15.8) ,8.3 73.2

TfR-F index 12.9 (28.1) ,5.6 46.4

Serum iron (mmol/l) 16.0 (15.7) 3.6–27.0 19.8

Serum transferrin (g/l) 2.2 (0.7) 2.0–3.6 2.5

Transferrin saturation (%) 41.4 (39.7) .16 37.4

TIBC (mmol/l) 41.5 (12.7) ,72 1.0

MCHC (g/dl) 32.9 (7.8) 32.0–36.8 43.3

MCV (fl)

,2 years 106.3 (81.4) 67–91 12.1

>2 years 117.7 (88.1) 73–89 21.5

MCHC, mean cell haemoglobin concentration; MCV, mean cell volume; sTfR, solubletransferrin receptor; TfR-F index, transferrin-ferritin index; TIBC, total iron bindingcapacity.

Table 3 Sensitivity and specificity of iron markers to identify children with iron stores deficiency usinginternationally accepted cut-off values and bone marrow iron as the ‘‘gold standard’’

True False

Sensitivity Specificity AccuracyPositives Negatives Positives Negatives

Ferritin 5 76 3 19 21 96 79

sTfR 35 52 87 4 90 37 49

TfR-F index 16 55 18 7 70 75 74

Serum iron 7 93 13 22 24 88 74

Serum transferrin 0 127 0 33 0 100 79

TransferrinSaturation

10 80 26 19 35 76 67

TIBC 0 127 0 33 0 100 79

MCHC 22 81 40 13 63 67 66

MCV

,2 years 1 71 8 22 4 90 71

>2 years 2 36 7 10 17 84 69

AUC, area under the curve; MCHC, mean cell haemoglobin concentration; MCV, mean cell volume; ROC, receiver operatingcharacteristics; sTfR, soluble transferrin receptor; TfR-F index, transferrin-ferritin index; TIBC, total iron binding capacity.

Original article

1104 J Clin Pathol 2009;62:1103–1106. doi:10.1136/jcp.2009.066498

a markedly lower sensitivity than specificity. Conversely, sTfRhad a high sensitivity of 90% and low specificity of 37%. Theaccuracy of a marker was highest for ferritin (79%) and lowestfor sTfR (49%).

ROC curves for defining the optimal cut for the identificationof deficiency of iron stores for ferritin, sTfR, TfR-F index, andother iron markers were constructed (fig 1). The area under thecurve (AUCROC) for each marker represented their performance(table 4). Ferritin, sTfR, TfR-F index and MCHC hadsignificantly higher AUCROC than 0.5, where 0.5 signified anineffective test. Serum iron and transferrin saturation wereexcluded from further analysis as they had non-significantAUCROC (0.62 and 0.60, respectively).

For the remaining iron markers, ferritin, sTfR, TfR-F indexand MCHC, new cut-off levels, with an optimal combination ofsensitivity and specificity, were determined from the ROCcurves. Table 5 shows the resultant sensitivity and specificity,and the percentage change from the original cut-off. The abilityof sTfR or ferritin to predict iron stores deficiency (accuracy)was similar and above 75% using the derived cut-offs of 273 mg/land 15.2 mg/ml, respectively. Although the sensitivity andspecificity for TfR-F index or MCHC (sensitivity 74%, 67%;and specificity 73%, 64%, respectively) were lower than thosefor ferritin and sTfR, they required smaller changes from theiroriginal to new cut-off levels.

DISCUSSIONIn the present study, the diagnostic efficiency of sTfR and a varietyof more conventional laboratory tests for the identification of

deficiency of iron stores or functional deficiency was evaluated.The results suggested that serum transferrin, TIBC and transferrinsaturation were of limited value in diagnosis of deficiency of ironstores as their corresponding AUCROC values did not provideacceptable sensitivity and specificity estimates. Most likely, thisrelates to the interference of inflammatory cytokines produced aspart of the acute phase response during an infection.11 12

The bone marrow iron smear was used as the ‘‘gold standard’’for the diagnosis of deficiency of iron stores. It has generallybeen considered the most reliable diagnostic test, but has thelimitations of being more invasive than peripheral blood ironmarkers.1 Furthermore, incorrect assessment of iron stores inbone marrow aspirates has been described.13

Ferritin has been widely used as an iron marker in individualswithout inflammatory conditions.6 Conversely, sTfR is con-sidered to reflect the degree of tissue iron need, and there isevidence that it is a good indicator of iron status when the ironstores are depleted.14 The reciprocal relationship between sTfRand ferritin describes a perfect log-linear relationship over awide range of normal and depleted iron stores states. Punnonenet al7 evaluated various possibilities of combinations of sTfR andferritin parameters, and concluded that use of the sTfR/logferritin ratio (TfR-F index) considerably improved diagnosticefficiency, even in settings with a high infection pressure.

In the present study, serum ferritin had a high specificity andsTfR a high sensitivity, which is consistent with previousstudies.15 Changing the current conventional cut-off of TfR-Findex (5.6) to its optimal cut-off (according to the ROCanalysis) of 5.3 has little effect on its diagnostic efficiency. Thisis in contrast with ferritin and sTfR which required a change of810% and 80%, respectively, to achieve maximal sensitivity andspecificity.

During a study in which healthy volunteers were seriallyphlebotomised to induce iron deficiency, MCHC was found tobe a sensitive early indicator of iron deficient erythropoiesis.16 17

In the present study, MCHC had a relatively good diagnosticefficiency. This is relatively good news since MCHC can bemeasured using a coulter counter, which is relatively cheap and

Figure 1 Received operating characteristic curves of soluble transferrinreceptor (sTfR), ferritin and transferrin-ferritin index (TfR-F index) in theidentification of iron stores deficiency. MCHC, mean cell haemoglobinconcentration.

Table 4 AUCROC values for biochemical iron markers to identify childrenwith deficiency of iron stores based on new cut-off values

AUCROC SE p Value

Ferritin 0.82 0.05 ,0.001

sTfR 0.80 0.04 ,0.001

TfR-F index 0.79 0.06 ,0.001

Serum iron 0.58 0.06 0.2

Transferrin saturation 0.57 0.06 0.3

TIBC 0.52 0.06 0.8

MCHC 0.68 0.06 0.001

AUC, area under the curve; MCHC, mean cell haemoglobin concentration; ROC,receiver operating characteristics; sTfR, soluble transferrin receptor; TfR-F index,transferrin-ferritin index; TIBC, total iron binding capacity.

Table 5 Ability of ferritin, sTfR, TfR-F index and MCHC to identify children with deficiency of iron storesbased on new cut-off values

Original cut-off New cut-off% change incut-off Sensitivity* Specificity* Accuracy*

Ferritin 30 mg/l 273 mg/l 810 75 76 76

sTfR 8.3 mg/ml 15.2 mg/ml 83 77 76 76

TfR-F index 5.6 5.3 25 74 73 73

MCHC 32.0 g/l 32.1 g/l 0.3 67 64 65

*Sensitivity, specificity or accuracy of new cut-off.MCHC, mean cell haemoglobin concentration; sTfR, soluble transferrin receptor; TfR-F index, transferrin-ferritin index.

Original article

J Clin Pathol 2009;62:1103–1106. doi:10.1136/jcp.2009.066498 1105

more widely available in resource-limited settings as comparedwith either ferritin or sTfR.

Results from the present study suggest that it is necessary tochange the cut-off limit for ferritin from 30 to 273 mg/l in orderto improve its diagnostic efficiency. This proposed increase isconsistent with other studies and probably reflects the effect ofthe acute phase response on ferritin levels.7 Witte et al18

developed a nomogram describing the relationship betweenferritin and CRP or erythrocyte sedimentation rate, to detect orexclude iron deficiency in patients with anaemia of inflamma-tion in order to minimise bone marrow examination.Unfortunately, when this nomogram, which corrected for theacute phase component of changes in ferritin, was applied inlater studies, it performed poorly.19

Results from the present study indicates that ferritin andsTfR are relatively good markers for detecting iron storesdeficiency provided that new cut-off values are applied.Combining these markers into the TfR-F index may prove tobe a much better tool to detect iron deficiency in children inthese high infection pressure areas. The cut-off values for theTfR-F index are robust and do not need to be adjusted forinflammation.

Our study is among the largest and most detailed investiga-tions into assessment of the iron status in children in thissetting and has generated a unique dataset. However, the factthat we did not do a healthy population-based study, butfocused on a subgroup of severely anaemic children, may limitgeneralisability of our findings, especially in view of the factthat these children also tend to have other inflammatoryconditions that may affect interpretation of ferritin levels.Obvious ethical issues, of performing bone marrow investiga-tions in mild and non-anaemic children, was the main reasonfor focusing on this study population. However, these findingsmay be a starting point and may provide an improvedknowledge of diagnostic criteria for iron status assessmentthat avoids the need to do a bone marrow aspiration, and be ofvalue for determining therapeutic practice. This is especially

important taken the recent observations that iron supplementa-tion to iron replete children may be fatal.3

Acknowledgements: We are indebted to the children and their guardians whoparticipated in this study and to the dedication of the Severe Anaemia Research Team.

Funding: The study was supported by a grant from the Gates Malaria Partnership(GMP Doctoral Fellowship to KP). Other contributions were from Numico and TerMeulen Foundation and the Wellcome Trust. The sponsor of the study had no role inthe study design, data collection, data analysis, data interpretations, or writing of thisreport.

Competing interests: None.

Ethics approval: Ethics approval was obtained.

Provenance and peer review: Not commissioned; externally peer reviewed.

REFERENCES1. Bain BJ. Bone marrow pathology, 3rd edn. Milan: Blackwell Science, 2001.2. Beguin Y. Soluble transferrin receptor for the evaluation of erythropoiesis and iron

status. Clin Chim Acta 2003;329:9–22.3. Sazawal S, Black RE, Ramsan M, et al. Effects of routine prophylactic

supplementation with iron and folic acid on admission to hospital and mortality inpreschool children in a high malaria transmission setting: community-based,randomised, placebo-controlled trial. Lancet 2006;367:133–43.

4. Calis J, Phiri K, Faragher B, et al. Aetiology of severe anemia in Malawian children.N Engl J Med 2008;358:888–99.

5. Molyneux EM, Tembo M, Kayira K, et al. The effect of HIV infection on paediatricbacterial meningitis in Blantyre, Malawi. Arch Dis Child 2003;88:1112–8.

6. WHO. Iron deficiency: assessment, prevention, and control. A guide for programmemanagers. Geneva: World Health Organization, 2001.

7. Punnonen K, Irjala K, Rajamaki A. Serum transferrin receptor and its ratio to serumferritin in the diagnosis of iron deficiency. Blood 1997;89:1052–7.

8. Phiri KS, Calis JCJ, Kachala D, et al. Improved method for assessing iron stores inthe bone marrow. J Clin Pathol 2009;62:1103–6.

9. Gabay C, Kushner I. Acute-phase proteins and other systemic responses toinflammation. N Engl J Med 1999;340:448–54.

10. van den Broek NR, Letsky EA, White SA, et al. Iron status in pregnant women:which measurements are valid? Br J Haematol 1998;103:817–24.

11. Beesley R, Filteau S, Tomkins A, et al. Impact of acute malaria on plasmaconcentrations of transferrin receptors. Trans R Soc Trop Med Hyg 2000;94:295–8.

12. Das BS, Thurnham DI, Das DB. Influence of malaria on markers of iron status inchildren: implications for interpreting iron status in malaria-endemic communities.Br J Nutr 1997;78:751–60.

13. Barron BA, Hoyer JD, Tefferi A. A bone marrow report of absent stainable iron is notdiagnostic of iron deficiency. Ann Hematol 2001;80:166–9.

14. Skikne BS, Flowers CH, Cook JD. Serum transferrin receptor: a quantitative measureof tissue iron deficiency. Blood 1990;75:1870–6.

15. Means RT Jr, Allen J, Sears DA, et al. Serum soluble transferrin receptor and theprediction of marrow aspirate iron results in a heterogeneous group of patients. ClinLab Haematol 1999;21:161–7.

16. Patton WN, Cave RJ, Harris RI. A study of changes in red cell volume andhaemoglobin concentration during phlebotomy induced iron deficiency and ironrepletion using the Technicon H1. Clin Lab Haematol 1991;13:153–61.

17. Bain BJ. Blood cells: a practical guide, 2nd edn. Oxford: Blackwell Science, 1995.18. Witte DL, Angstadt DS, Davis SH, et al. Predicting bone marrow iron stores in

anemic patients in a community hospital using ferritin and erythrocyte sedimentationrate. Am J Clin Pathol 1988;90:85–7.

19. Coenen JL, van Dieijen-Visser MP, van Pelt J, et al. Measurements of serum ferritinused to predict concentrations of iron in bone marrow in anemia of chronic disease.Clin Chem 1991;37:560–3.

Take-home messages

c There is a need to properly diagnose true iron deficiency dueto the probable increased risk of adverse effects associatedwith iron supplementation in iron replete individuals.

c Transferrin-ferritin index is probably the most useful and robustiron marker that best predicts bone marrow iron status.

c Mean cell haemoglobin concentration may be an acceptablealternative screening test in resource-poor settings.

Original article

1106 J Clin Pathol 2009;62:1103–1106. doi:10.1136/jcp.2009.066498